Selenium is a browser automation tool used primarily for testing web applications. However when Selenium is used in a CI pipeline to test applications, there is often contention around the use of Selenium resources. This example shows you how to deploy Selenium to Kubernetes in a scalable fashion.
### Prerequisites
This example assumes you have a working Kubernetes cluster and a properly configured kubectl client. See the [Getting Started Guides](../../docs/getting-started-guides/) for details.
Google Container Engine is also a quick way to get Kubernetes up and running: https://cloud.google.com/container-engine/
Your cluster must have 4 CPU and 6 GB of RAM to complete the example up to the scaling portion.
### Deploy Selenium Grid Hub:
We will be using Selenium Grid Hub to make our Selenium install scalable via a master/worker model. The Selenium Hub is the master, and the Selenium Nodes are the workers(not to be confused with Kubernetes nodes). We only need one hub, but we're using a replication controller to ensure that the hub is always running:
Let's verify our deployment of Selenium hub by connecting to the web console.
#### Kubernetes Nodes Reachable
If your Kubernetes nodes are reachable from your network, you can verify the hub by hitting it on the nodeport. You can retrieve the nodeport by typing `kubectl describe svc selenium-hub`, however the snippet below automates that by using kubectl's template functionality:
```console
export NODEPORT=`kubectl get svc --selector='app=selenium-hub' --output=template --template="{{ with index .items 0}}{{with index .spec.ports 0 }}{{.nodePort}}{{end}}{{end}}"`
export NODE=`kubectl get nodes --output=template --template="{{with index .items 0 }}{{.metadata.name}}{{end}}"`
curl http://$NODE:$NODEPORT
```
#### Kubernetes Nodes Unreachable
If you cannot reach your Kubernetes nodes from your network, you can proxy via kubectl.
```console
export PODNAME=`kubectl get pods --selector="app=selenium-hub" --output=template --template="{{with index .items 0}}{{.metadata.name}}{{end}}"`
kubectl port-forward --pod=$PODNAME 4444:4444
```
In a seperate terminal, you can now check the status.
```console
curl http://localhost:4444
```
#### Using Google Container Engine
If you are using Google Container Engine, you can expose your hub via the internet. This is a bad idea for many reasons, but you can do it as follows:
Then wait a few minutes, eventually your new `selenium-hub-external` service will be assigned a load balanced IP from gcloud. Once `kubectl get svc selenium-hub-external` shows two IPs, run this snippet.
```console
export INTERNET_IP=`kubectl get svc --selector="app=selenium-hub,external=true" --output=template --template="{{with index .items 0}}{{with index .status.loadBalancer.ingress 0}}{{.ip}}{{end}}{{end}}"`
curl http://$INTERNET_IP:4444/
```
You should now be able to hit `$INTERNET_IP` via your web browser, and so can everyone else on the Internet!
You now have 10 Firefox and 10 Chrome nodes, happy Seleniuming!
### Debugging
Sometimes it is neccessary to check on a hung test. Each pod is running VNC. To check on one of the browser nodes via VNC, it's reccomended that you proxy, since we don't want to expose a service for every pod, and the containers have a weak VNC password. Replace POD_NAME with the name of the pod you want to connect to.
```console
kubectl port-forward --pod=POD_NAME 5900:5900
```
Then connect to localhost:5900 with your VNC client using the password "secret"
